LEARN. PRACTICE. GROW.Identify the decision and its mandate
Before requesting option approval, distinguish architecture recommendation, risk acceptance and change authorization. These may belong to different owners. In the example, the team may recommend a pattern but cannot accept residual risk for another forum. Its work should enable that decision: alternatives, consequences, affected requirements and pending matters. Excluding every option requiring another authority may also be premature. Confirm the mandate and coordinate the decision within the deadline without presenting a technical recommendation as authorization to execute.
Turn disagreement into investigation
APS predicts more overnight interventions; delivery predicts fewer. Averaging the estimates is not an observed result. First agree what counts as an intervention, which period and volume matter, and what conditions represent operation. Then collect comparable evidence. Preparation also includes commitments: minutes assigning weekend coverage do not establish capacity from an absent team. Where the process requires explicit acceptance, retain a pending commitment until owner and availability are confirmed. Silence does not satisfy that condition or supply the missing operational capacity.
Compare value over the agreed horizon
Monthly savings can hide when benefits start. In a twelve-month model, A saves €12k per month only in months 7 through 12: six months and €72k. B saves €8k from the first month: €96k. With equivalent risks and other costs, B better serves the objective over that horizon. A longer period may change the comparison; do not change it merely to favor a proposal. Record horizon, benefit start and exclusions. These values are teaching assumptions rather than validated financial forecasts.
Separate past cost from future choice
The prototype consumed €90k that cannot be recovered. The next decision should compare costs and benefits that can still change: A costs €60k and brings €100k, leaving €40k net; B costs €30k and brings €80k, leaving €50k net. Past value remains in financial reporting and learning but does not make A mandatory. Use this simple comparison only under the defined assumptions: same horizon, equivalent risk and no discounting. If exit costs or different obligations exist, include them before recommending.
Use discovery to reduce decisive uncertainty
An undocumented connector can govern the entire vision. If ten days remain before committing to a deadline, producing more diagrams of known components may not help. Propose time-bounded discovery with the question to answer, expected evidence and next decision. Analogy with a simple connector may start an estimate but does not guarantee equivalence. The result may justify an option, revise timing or retain explicit uncertainty. Do not declare risk resolved merely because the discovery activity finished within its allotted window.
Exercise: close data and outcome contracts
Spend fifteen minutes representing two sources: a position snapshot at 18:00 and live data at 18:05. Write why transporting both does not necessarily produce a coherent position at 18:00. Then analyze an API that accepts an instruction before losing its response. The client sees a timeout, but that does not prove rejection. Define operation identity, retry treatment and outcome lookup. Use the remaining minutes to prepare questions for business and application owners. The proposal should identify missing decisions without inventing semantics or authority absent from the case.
Prepare a decision note with a twelve-month horizon, A=€72k and B=€96k. Add the incremental A=€40k/B=€50k calculation as a separate exercise; do not add option models belonging to different decisions.
Common pitfalls
Presenting recommendation as authorization; treating silence as commitment; comparing savings with different starts; justifying future work only by past spending; interpreting timeout as proof of rejection.
Related topics: Vision and feasibility · Stakeholders and commitments · Contracts and temporal data
A useful decision connects objective, authority, horizon and evidence. Calculation organizes assumptions, coordination confirms commitments, and architecture makes explicit the conditions under which an option can produce the outcome.
Reference: Practitioner competency-to-role mapping · OGEA-102; TOGAF Standard, 10th Edition